CN107661087B

CN107661087B - Medical imaging apparatus and method for imaging of photosensitive objects such as biological tissue

Info

Publication number: CN107661087B
Application number: CN201710562149.2A
Authority: CN
Inventors: 乔治·塞梅利斯
Original assignee: Leica Instruments Singapore Pte Ltd
Current assignee: Leica Instruments Singapore Pte Ltd
Priority date: 2016-07-29
Filing date: 2017-07-11
Publication date: 2019-12-31
Anticipated expiration: 2037-07-11
Also published as: CN107661087A; US20180033141A1; JP6400159B2; US11501427B2; EP3275361A1; EP3275361B1; JP2018029954A

Abstract

The invention relates to a medical imaging apparatus (1) and a method for imaging a photosensitive object (2), such as biological tissue (3), in particular an eye (4). The aim of the invention is to present a device and a method that allow the display of objects (2) in spectral bands (30) of the visible range (37) to which the objects are sensitive and therefore cannot be illuminated. This object is achieved by a medical imaging apparatus (1) comprising a camera for capturing an input image (10). Thus, the spectral features (36) of the object are represented in the input image by an input set (15) of spectral bands (16). The apparatus produces an output image (22) in which spectral features (60) in the visible light range (37) are represented by an output set (29) of spectral bands (30). By replacing at least one input-only spectral band (38) in the input set (15) with at least one output-only spectral band (39) in the output set (29), it is possible to capture an input image (10) in spectral bands (16) in which the object (2) is insensitive or hardly sensitive, such as the invisible light range (18), and to display the object (2) in spectral bands (30) in which the object may be too sensitive to illumination (40). Replacing at least one input-only spectral band (38) with at least one output-only spectral band (39) is accomplished using the calibration data (56) and the image processor (31).

Description

Medical imaging apparatus and method for imaging of photosensitive objects such as biological tissue

Technical Field

The present invention relates to a medical imaging apparatus and method for imaging of a light-sensitive object having spectral characteristics, such as biological tissue, in particular, light-sensitive tissue of a living body, such as an eye. In particular, the medical imaging apparatus and method are ophthalmic apparatus and method, respectively.

Background

Viewing and imaging of objects (e.g., living tissue) is complicated if the object is photosensitive. When illuminated, photosensitivity may cause tissue to react to it. This response may cover the characteristics of the tissue that needs to be observed to make a diagnosis. In some cases, the illumination may even damage the tissue.

In ophthalmology, sensitivity of the sensitive tissues in the eye causes discomfort, a high risk of injury and additional difficulties to the patient due to the reflex response of the eye to light.

To avoid these problems, color cameras are used which produce acceptable image quality in low light conditions. However, such cameras are expensive and bulky, and require expensive and large high quality optical systems. The large aperture of the optical system required to capture as much incident light as possible only results in a limited depth of field.

Disclosure of Invention

The present invention aims to provide a medical imaging apparatus and method which overcomes these problems in the examination of living photosensitive tissue.

This object is achieved according to the invention with a medical imaging apparatus of the kind mentioned initially, wherein the medical imaging apparatus comprises a camera for capturing an input image in which spectral features of the object are represented by an input set of spectral bands, an output interface for delivering the output image, for example to a peripheral device, wherein in the output image the spectral features in the visible light range are represented by an output set of spectral bands, wherein at least one input-only spectral band of the input set is replaced by at least one output-only spectral band of the output set, the medical imaging apparatus further comprising a storage component in which calibration data are stored for mapping the input set to the output set of spectral bands, and an image processor for determining an intensity in the at least one output-only spectral band in dependence on the input set and the calibration data.

The medical imaging method according to the invention comprises the steps of: acquiring an input image of an input set of spectral bands, transmitting an output image consisting of an output set of spectral bands, the output set representing spectral features in the spectral bands of the output set, converting the input image into the output image by replacing at least one input-only spectral band of the input set with at least one output-only spectral band of the output set and adapting the intensity in the at least one output-only spectral band to the spectral features using previously stored calibration data.

The present invention uses information included in an input set of spectral bands to determine spectral features of an object and renders the spectral features in an output image using an output set of spectral bands different from the input set. This allows recording images in spectral bands (e.g. IR or NIR range) where the object, in particular the light sensitive tissue, does not have any photosensitivity. Using the calibration data, the "true" spectral features of the object are rendered using spectral bands of the output set within a different spectral range (e.g., visible range) than the spectral range in which the input image was captured.

If the output image is displayed on a display, the viewer can see the very same color as viewing tissue in the output set of spectral bands. Thus, the imaging is done in a spectral range that is not dangerous to the tissue, e.g. not seen by the human eye, while the display is done in the visible range to allow the viewer to correctly identify the tissue by its natural color.

The method and device according to the invention have the advantage of reducing patient discomfort and significantly reducing the risk of damaging tissue. The method and apparatus of the present invention allow for significantly prolonged examination which helps to avoid errors in diagnosis and treatment.

The apparatus and methods may also be used in non-medical applications. For example, in military applications, it may be possible to observe an object in ambient light having only a limited number of spectral bands of sufficient intensity. Use of the present invention allows objects to be displayed in their natural color.

The method and the device according to the invention can also be improved by the following features, which are independent of one another in terms of their respective technical effect and which can be combined in any desired manner.

For example, preferably at least two spectral bands of the input set (hereinafter referred to as input spectral bands) are in the longer invisible wavelengths, in particular in the IR light range. Especially in ophthalmology, tissues are significantly less sensitive to longer invisible wavelengths (e.g. UV) than to shorter invisible wavelengths. However, shorter wavelengths in the blue or UV spectral band may also be used, especially for studying surface tissue layers, due to the low penetration of shorter wavelengths of light.

The input and output images are portions of a time series of input and output images that are captured and/or displayed, the input and output images being captured at a frame rate that is higher than the flicker fusion rate so that any changes in tissue during the observation period can be smoothly reproduced in the output images.

The input image may comprise input pixels. The output image may include output pixels. The input image may be a color image. The output image may be a color image.

According to one aspect, each input pixel represents a location in the field of view of the camera. Preferably, the output image and the input image are identical and have the same orientation. They may have the same resolution, pixel count, size and/or aspect ratio. Preferably, the input pixels in the input image and the output pixels in the output image represent the same or at least coinciding positions in the field of view or the same region, respectively, of the observed tissue.

The conversion from the input set to the output set is preferably performed on a pixel basis: for each pixel in the input image, a corresponding intensity in only the output spectral band is calculated depending on the intensity in the input spectral band and the calibration data.

According to a further advantageous embodiment of the invention, the medical imaging apparatus may comprise an illumination system for illuminating the object with light consisting of an illumination set of spectral bands. The illumination light spectral band may be a spectral band at least partially in the invisible light range, in particular at least in the input spectral band. Preferably, only the output spectral band is attenuated or absent from the illumination set. Illuminating objects in the input spectral band guarantees good quality of the input image in terms of depth of field and spatial resolution without the need for spatial low-light optics or cameras. There is no risk of damaging tissue or causing discomfort to the patient in the input spectral band. Furthermore, in an ophthalmic examination, by shrinking the iris, the eye will not react to illumination in the invisible light range, thus allowing a simpler examination.

In other applications, not medical, such illumination may prevent detection of the illumination source.

Preferably, the illumination is limited to the invisible light range. For example, illumination filtering may be provided that limits illumination of tissue to the invisible light range.

Since only illumination in the input spectral band is required, it is sufficient to limit the illumination to the input spectral band. Thus, the illumination filtering may comprise a band-pass filter and/or a band-stop filtering system that is transmissive in the input spectral band.

According to another aspect of the invention, the illumination may preferably comprise discrete illumination spectral bands matching the input spectral bands of the multispectral camera. The input spectral band and the illumination spectral band may be comprised of light having a wavelength of at least 550 nm. In particular, the illumination system may comprise IR and/or NIR light sources.

To avoid irritating the eye or any other type of interference with objects, in one embodiment most, if not all, of the input-only spectral bands lie outside the visible range, particularly in the IR or NIR range, whereas most, if not all, of the output-only spectral bands lie within the visible range. The input and/or output spectral bands may be discrete and/or have substantial (substitional) overlap. Most, if not all, of the spectral bands in the input set may be located in the non-visible light range to collect as much information as possible about the spectral characteristics of the object in the insensitive spectral bands. For optimal reproduction of the color of the object, most, if not all, of the spectral bands in the output set may be in the visible range.

The camera may be an imaging spectrometer, in particular a multi-spectral camera or a hyperspectral camera. The more spectral bands that are used, the better the spatial resolution of the input spectral features and the more accurate the assignment of the spectral features of the object to the output set. For some applications, a standard color camera, for example, operating the RGB color space may be sufficient.

In one embodiment, the output set may preferably comprise only three discrete spectral bands and may in particular consist of a tristimulus space (e.g. RGB). Such a color system is sufficient for the output pixels to faithfully reproduce the true color of the tissue to a human observer.

According to another aspect of the invention, an image processor may be used to convert the input set to the RGB space. Here, the set of outputs includes R-, G-, and B-spectral bands. In theory, this is an ill-posed problem and cannot be solved because the spectral features in the input set alone do not contain enough information to determine the color appearance of tissue in spectral bands outside of the input set. However, prior knowledge of the optical (e.g., reflective) properties of biological tissue may provide the additional information needed to calculate this transformation. This information is included in the correction data.

The image processor may be configured to classify the tissue represented by the input pixels according to the input spectral features by spectral fitting. The image processor may comprise a spectral analysis module for performing a spectral fit on the input spectral features and for calculating tissue reference values from the spectral fit, wherein the image processor is for determining the output spectral features from the biological reference values. Thus, the spectral analysis module may perform a classification of the tissue represented by the input pixels. Classification may be accomplished by assigning tissue reference values to the respective input spectral features. This value is then characteristic of a particular kind of tissue (e.g. tissue with a specific concentration of a bio-pigment). By determining the output spectral characteristics from the tissue reference values, the visible appearance of the tissue in the visible range can be matched to tissue having specific input spectral characteristics in the invisible range. The tissue reference value represents the color of the tissue in the visible range.

The calibration data may include a spectral library. The spectral library may be indexed using the tissue reference values. Each of the plurality of output spectral features may be assigned to a unique respective tissue reference value.

The calibration data may include parameters of a transfer function that maps the input set to the output set. Such a transfer function may be determined empirically or analytically, for example, based on a light-tissue interaction model.

The calibration data may include parameters of a mechanical test system, such as an artificial network trained using a set of model inputs and outputs.

The organizational reference value may be any from a single bit to a bit field, which in a particular example may be a single number or an array of numbers. It may represent one or more quantities, such as exponents of an exponential fourier series, or RGB values. The tissue reference value may also be an array of values indicative of the concentration of the bio-pigments contributing to and/or causing the input spectral feature. Such an array can be used to analyze output spectral features in a visible light image from model output spectral features in a spectral library, where each output spectral feature represents a color of a component biological pigment in the visible light range.

If the concentration of the biological pigment in the tissue is automatically determined from the input spectral features of the input pixels, for example by spectral fitting as mentioned above, the output spectral features may be composed of model output spectral features according to this concentration. The biological pigments whose concentrations are determined and/or whose model output spectra are stored in the spectral library include at least one of oxyhemoglobin, deoxyhemoglobin and at least one melanin pigment, such as eumelanin. Preferably, the number of input spectral bands is at least as large as the number of biological pigments represented in the spectral library.

Instead of such a composition of output spectral features from a plurality of model output spectral features, it is also possible to directly find the output spectral features in a differently constructed spectral library.

In another embodiment, to perform the spectral fitting, the storage means may comprise a representation of stored model input spectral features of certain biological pigments, for example as described above, for spectral fitting of the input pixel features of the input pixels.

The image processor and any modules comprised by the image processor may be constituted by hardware and/or software.

The present invention may also relate to a non-transitory computer storage medium storing a program that causes a computer to execute the medical imaging method in one of the above-described embodiments. The non-transitory computer storage medium may be part of a medical imaging apparatus.

Drawings

In the following, the method and the device according to the invention are explained in more detail with reference to the enclosed drawings, in which exemplary embodiments are shown.

In the figures, the same reference numerals are used for elements corresponding to each other in their functions and/or structures.

In accordance with the description of the various aspects and embodiments, elements shown in the figures may be omitted if the technical effect of such elements is not required for a particular application. Vice versa, elements not shown or described with reference to the figures, but described above, may be added if the technical effect of a particular element in a particular application is beneficial.

In the figure:

fig. 1 shows a schematic representation of a medical imaging apparatus according to the invention;

fig. 2 shows a schematic reproduction of the spectrum of the biological pigment.

Fig. 3 shows a schematic representation of a medical imaging method according to the invention.

Reference numerals

1 medical imaging apparatus

2 object

3 organization

4 eyes

6 ophthalmic imaging device

8 multispectral camera

10 input image

12 field of view

14 input pixel

Input set of 15 spectral bands

16 input spectral bands

18 range of invisible light

20 output interface

22 output image

24 display

26 output pixel

28 position of input pixel in field of view

Output collection of 29 spectral bands

30 output spectral band

31 image processor

32 input spectral features recorded by a camera

34 output spectral characteristics of the output in the display

36 true spectral features of the object

37 visible light range

38 input spectral bands only

39 output spectral bands only

40 illumination system

42 light generated by the illumination system

Illumination collection of 44 spectral bands

46 illumination spectral band

48 filtering system

50 spectral analysis module

52 tissue reference value

54 storage element

56 calibration data, e.g. spectral library

58 model input spectral features

60 model output spectral features

62 optical system

64 light path of the illumination system

66 optical path of camera

68 field of illumination

70 Beam splitter arrangement

72 observer

74 optical path of observer

75-79 model input and output spectral features

80 sample input spectral features

82 sample output spectral features

90 image acquisition step

Fitting of 92 spectra

94 tissue constituent determination

96 output image generation

98 image output step

NIR₁，NIR_nMultispectral data in each of n input spectral bands

BP₁，BP_mConcentration of each of m biochrome determined by spectral fitting

Detailed Description

First, an example of a medical imaging apparatus 1 is described with reference to fig. 1.

The medical imaging apparatus 1 is used for imaging an object 2, for example biological tissue 3, in particular a light sensitive area of tissue, for example an eye 4. In particular, the medical imaging apparatus 1 is used as an ophthalmic imaging device 6 in ophthalmology.

The medical imaging apparatus 1 comprises a camera 8 for capturing an input image 10 of the tissue 3 located in a field of view 12. The camera 8 may be a color camera (e.g., an RGB camera) or an imaging spectrometer (e.g., a multispectral or hyperspectral camera).

Each input image 10 comprises input pixels 14. Each pixel 14 represents a region of the field of view. Each input image 10 and each input pixel 14 respectively contains an input set 15 of at least spectral bands 16.

The camera 8 is sensitive to at least two discrete input spectral bands 16 in the invisible light range 18. Invisible light consists of portions of the electromagnetic spectrum not visible to the human eye. Thus, the spectral band of the invisible light range 18 corresponds to wavelengths less than about 390nm or greater than about 700 nm. Preferably, the at least two spectral bands are in the infrared or near-infrared and thus include wavelengths greater than 700nm to about 1 mm.

The medical imaging apparatus 1 also preferably comprises a digital output interface 20 for preferably digitally transmitting an output image 22 to a peripheral device 24, such as a display. The display 24 displays the output image 22 in the visible range having a wavelength between about 390nm and about 700 nm.

The output image 22 is a color image and includes output pixels 26. The output image 22 may be in a standard color space form and may be, for example, an RGB image.

Preferably, the input image 10 and the output image 22 are consistent (consequent) and adjusted (corrected). Preferably, the input pixels 14 and the output pixels 26 are coincident. Thus, the position of the output pixel 26 in the output image 22 corresponds to the position 28 of the respective input pixel 14 in the field of view 24. Each input pixel 14 contains spectral information of the corresponding location 28 and thus of the tissue 3.

In particular, each output image 22 or each output pixel 26 in an output image 22 may be made up of an output set 29 of spectral bands 30, the output set 29 of spectral bands 30 being referred to below as output spectral bands. At least one spectral band of the input set 15 and the output set 29 is different, preferably the at least one spectral band is in the visible range.

Preferably, the spatial resolution of the output image 22 corresponds to, or is smaller than, the spatial resolution of the input image 10. Thus, the size and position of the portions of the object 2 represented by the input and output pixels 14, 26 match each other in the input and output images 10, 22. Alternatively, the output image 22 may contain fewer or more output pixels than the input image 14. In the former case, multiple input pixels 14 may be combined into an output pixel 26. In the latter case, the input pixel 14 may be copied onto a plurality of output pixels 26.

The medical imaging apparatus 1 may further comprise an image processor 31 for generating the output pixels 26 from the input pixels 14 and for replacing input spectral features 32 of the input pixels 14 with output spectral features 34 in the output pixels 26, wherein the input spectral features are represented by the input set 15 and the output spectral features are represented by the output set 29. The output pixels 26 that include the replacement output spectral features 34 correspond to the positions and sizes of the input pixels 14 in the input image 10 in terms of positions and sizes in the output image 22. The input spectral features 32 correspond to the intensity distribution in the input spectral band 16 due to the change in reflectance across the spectrum at each location 28.

Preferably, the output spectral features 34 of the output pixels 26, represented by the spectral bands 30 of the output set 29, correspond at least to an approximation of the spectral features 36 of the tissue 3 in the visible range 37 of light at the location 28 of the input pixel 14 in the field of view 12. The output image 22 thus contains a computed rendition of the visible color of the tissue 3 in the field of view 12, and thus those that would be visible to an observer when the input image 10 was captured within the visible light range 37. In each pixel of the output image 22, the input set 15 is at least partially replaced by the output set 29.

In particular, input set 15 includes at least one input-only spectral band 38 that is not included in output set 19. Preferably, only the input spectral bands lie within the invisible light range 18. In the output set 29, at least one output-only spectral band 39 is preferably contained within the visible range 37 and not in the input set 15. The replacement of at least one input-only spectral band 38 by at least one output-only spectral band 39 is done by the image processor 31. The image processor 31 may be implemented in any combination of hardware and software or alone in hardware or software.

The medical imaging apparatus 1 may further comprise an illumination system 40 for illuminating the tissue 3 in the invisible light range 18. The illumination system 40 is for producing light 42 comprising at least the input spectral band 16. Light 42 may be comprised of an illumination collection 44 of illumination spectral bands 46 that includes at least output spectral band 30. In one embodiment, the illumination set 44 may correspond to the input set 5.

A filtering system 48 may be provided for the illumination 40, and the filtering system 48 may block light 42 in the visible range and/or limit the light 42 to the input spectral band 16. In particular, the illumination system 40 may comprise one or more IR and/or NIR light sources.

Preferably, the image processor 31 comprises a spectral analysis module 50, the spectral analysis module 50 being configured to perform a spectral fit to the input spectral features 32 and to calculate tissue reference values 52 from the spectral fit. The image processor 31 may be used to retrieve or calculate the output spectral features 34, i.e. the intensities in at least one output spectral band 39 only, independently of the tissue reference values 52. The tissue reference value represents the color of the tissue 3 in the visible range. In particular, the tissue reference value 52 may represent the content and/or concentration of a particular bio-pigment or chromophore at the location 28 of the tissue 3 mapped to the input pixel 14. The organizational reference value is a numerical value, which may be a bit or a field of bits. The tissue reference value may comprise one or more quantities, such as a weighting factor, an exponential or an exponential of a fourier series, or an RGB value.

The storage component 54 of the medical imaging apparatus 1 may contain calibration data 56, the calibration data 56 being used for mapping the input set 15 to the output set 29 and thus for reproducing the true spectral features 36 of the object 2 in the output image 22. The calibration data 56 may represent parameters that allow the transfer function of the output set 29 to be calculated from the input set 15. The calibration data may represent parameters of a learning function or neural network that has been trained for calibration purposes using the input set 15 and the matched output set 29. The calibration data 56 may be representative of various spectral features in the form of at least one spectral library 56. This representation may be in the form of a numerical value corresponding to the intensity in the respective spectral band of the color space defined by the input set 15 and the output set 29.

In one embodiment, the storage component 54 may store representations of model input spectral features 58 of various biological pigments in the spectral library 56. The model input spectral features 58 may be limited to the input spectral band 16. The model input spectral features 58 may be used to spectrally fit the input spectral features 32 of the input pixels 14 represented by the input set 15. In spectral fitting, the algorithm may automatically select a set of model input spectral features 58, the combination of model input spectral features 58 resulting in the closest approximation of a given input spectral feature 32 of the input pixel 14. This manipulation may be performed for all input pixels of the input image 10. The particular linear or non-linear combination of model input spectral features 58 in a given input spectral feature indicates which types of biological pigments are present in the location 28 represented by the respective input pixel 14. The weight of a particular model input spectral feature 58 within the combination of model spectral features 58 required to approximate the input spectral feature 32 is indicative of the concentration of the respective biological pigment in the location 28.

Alternatively or additionally, the spectral library 56 may include model output spectral features 60. The model output spectral features may be limited to at least one output-only spectral band 39. For each input spectral feature 32, a matching output spectral feature 60 is retrieved, for example, by using the tissue reference value 52. The intensity in at least one output-only spectral band 39 corresponding to the matched model input features 59 is determined from the model output spectral features 60.

In another embodiment, the model output spectral features 60 stored in the spectral library 56 correspond to output spectral features 60 of different biochips. In such embodiments, the output spectral features 34 of the output pixels 26 are constituted, via the image processor 31, by respective model output spectral features 60 of the bio-pigments that have been found to contribute to the input spectral features 32. The latter embodiment is explained in more detail below.

The output spectral features 34 of the output pixels 26 may be comprised of a combination of model output spectral features 60. The model output spectral features 60 and the model input spectral features 58 may both be assigned to a particular biological pigment. The model output spectral features represent spectral features 36 of the biological pigments in the visible light range 37, in particular in the color space of the output image 22. The model input spectral feature 32 corresponds to a spectral representation of the bio-pigment in the non-visible range 18, in particular in the input spectral band 16.

The output spectral features 34 in the output image 22 may comprise combinations of the pattern output spectral features 60, the combinations of the output spectral features 60 together forming the output spectral features 34 of the output pixels 26. The weight of each model output spectral feature 60 in the output spectral features 34 is the weight of the corresponding model input spectral feature 58 in the input spectral features 32. The material composition of the tissue 3 at the location 28 of the input pixel 14 calculated during the spectral fitting process is used to construct the output spectral feature 34 in the visible range from the respective model output spectral features 60 of the component bio-pigments.

The medical imaging apparatus 1 may further comprise an optical system 62, the optical system 62 being located in an optical path 64 of the light 42 of the illumination system 40, an optical path 66 of the multispectral camera 8. Thus, the optical paths 64 and 66 are aligned and the illumination field determined by the illuminator with an illumination intensity above a particular threshold matches the field of view 12.

A beam splitter arrangement 70 may be used to split the optical paths 64 and 66.

A viewer 72, such as an eyepiece or binoculars, may also be provided to allow direct inspection of the field of view 12 by the optical system 62. Viewer 72 may have its own optical path 74 separated from optical paths 64 and 66 via beam splitter arrangement 70. The display 24 or an additional display may be arranged between the viewer 72 and the beamsplitter arrangement 70 in the light path 74.

The medical imaging apparatus 1 and the processing thereof as described above are also applied when the input image 10 and the output image 22 are three-dimensional.

In fig. 2, model input spectral features 58 in the invisible light range 18 and model output spectral features 60 in the visible light range 37 are shown for different biological pigments.

In fig. 2, line 75 shows the model input and output spectral features 58, 60 of deoxyhemoglobin, line 76 shows the model input and output spectral features of oxyhemoglobin, line 77 shows the model input and output spectral features of eumelanin, line 78 shows the model input and output spectral features of blood with 80% oxygen saturation, and line 79 shows the model input and output spectral features of adipose tissue. The model input and output spectral features 75-79 are in the visible range 37 and the invisible range 18, respectively.

The model output characteristics may be derived (particularly calculated) by using the calibration data 56 (e.g., a spectral library). Other forms of calibration data are transfer functions that map input spectral features to output spectral features, or parameters of self-learning computational structures (e.g. neural networks) that are calibrated prior to use of the apparatus 1.

In fig. 2, the model input and output spectral features form a sample spectral library 56 for the indicated biological pigment. It can be seen that the model output spectral feature 60 is merely a continuation or completion of the model input spectral feature 58 in the visible range 37 in the invisible range 18.

The model input and output spectral features 75-79 are schematic representations on arbitrary scales. The input spectral feature 32 of a given input pixel in the input spectral band 30 is illustratively indicated as line 80. In spectral fitting, a weighted sum of the model input spectral features 58 is calculated, for example by minimizing the minimum mean square error of the input spectral feature 30, which best approximates the sample input spectral features 80. The weights of the various model input spectral features 58 are then used to calculate the sample output spectral features 82 in the output spectral band 30. The weights of the individual model spectral features 54 are simply applied as applied to the model output spectral features 60.

As can be further seen from fig. 2, a model input spectral feature 58 can be determined from the input spectral bands 16 of the input set 15. Once the model input spectral features are identified (or objects or tissues are classified), the intensity in at least one output-only spectral band 39 can be obtained from the corresponding mode output features 60 (or classifications). The intensity in the spectral band 30 of the output set 29 then corresponds to the color of the object 2.

As can be seen from fig. 2, at least one spectral band can be shared by the input and output sets 15, 29.

Referring to fig. 3, a medical imaging method is briefly described.

In the image acquisitionMultispectral data NIR is obtained by the camera 8 for each input pixel 14 in the input image 10 in each of the n input spectral bands 16, taking 90₁To the NIR_n. The number n of input spectral bands 16 corresponds at least to the number m of bio-pigments, which are responsible for the color of the tissue 3 in the visible range.

By spectral fitting 92, the composition of the model input spectral features from the spectral library 56 is calculated that best matches the multispectral data NIR₁To the NIR_n. From the spectral fit 92, the bio-pigment BP in the tissue 3 at the position 28 is determined₁To BP_mThe concentration of (c).

In output image generation 96, output spectral features 34 are calculated for each output pixel 26 corresponding to an input pixel 14, with respect to a location 28 (FIG. 1) in the field of view 12. The output spectral feature 34 at each output pixel 26 is determined by the presence of the bio-pigment BP at the location 28 as determined by the spectral fit 92₁To BP_mOf the respective model output spectral features 60. The weight of each of the output spectral features 60 in the output spectral features 34 corresponds to the weight of the model input spectral feature 58 in the input spectral feature 32. As with the model input spectral features 58, the model output spectral features 60 are looked up in the spectral library.

Finally, in step 98, the output image 22 comprising the output pixels 26 is output in a color space form (e.g., RGB).

Claims

1. A medical imaging apparatus (1) for imaging of a light sensitive object (2) for imaging of biological tissue (3), the light sensitive object having a spectral feature (36), the medical imaging apparatus (1) comprising:

a camera (8) for capturing an input image (10) in which the spectral features (36) of the light-sensitive object (2) are represented by an input set (15) of input spectral bands (16),

an output interface (20) for transmitting an output image (22),

wherein in the output image (22) spectral features (60) within the visible light range (37) are represented by an output set (29) of output spectral bands (30),

wherein at least one input-only spectral band (38) of the input set (15) is replaced by at least one output-only spectral band (39) of the output set (29),

the medical imaging apparatus (1) further comprises:

a storage component (54) in which calibration data (56) is stored in the storage component (54) for mapping the set of inputs (15) to the set of outputs (29), an

An image processor (31) for determining intensities in the at least one output-only spectral band (39) from the set of inputs (15) and the calibration data (56).

2. The medical imaging apparatus (1) as defined in claim 1, further comprising an illumination system (40), the illumination system (40) being for illuminating the light-sensitive object (2) with light (42) consisting of an illumination set (44) of illumination light bands (46), wherein the at least one output-only spectral band (39) is attenuated or not included compared to the other output spectral bands (30).

3. The medical imaging apparatus (1) as defined in claim 1, wherein the at least one input-only spectral band (38) is located within the invisible light range (18).

4. The medical imaging apparatus (1) according to claim 2, wherein the illumination set (44) comprises at least one illumination spectral band (46) in the NIR range.

5. The medical imaging device (1) according to any one of claims 1 to 4, wherein a majority of the input spectral bands (16) in the input set (15) lie within the invisible light range (18) and a majority of the output spectral bands (30) in the output set (29) lie within the visible light range (37).

6. The medical imaging apparatus (1) according to any one of claims 1 to 4, wherein the storage component (54) comprises representations of spectral features (36) in at least the input spectral bands (16) of the input set (15) and the output spectral bands (30) of the output set (29), the representations of spectral features (36) corresponding to intensities of the respective spectral bands of the color space defined at the input set (15) and the output set (29).

7. The medical imaging apparatus (1) according to claim 6, wherein the storage component (54) comprises a bio-pigment (BP)₁To BP_m) For spectral fitting (92) of an input set (15) of input images (10).

8. The medical imaging apparatus (1) according to any one of claims 1 to 4, wherein the medical imaging apparatus (1) comprises a spectral library (56) comprising a set of model output spectral features (60), wherein the output set (29) comprises a combination of at least a subset of the model output spectral features (60) in the spectral library (56).

9. The medical imaging apparatus (1) according to claim 2, wherein at least one of the input set (15) and the illumination set (44) represents light (42) having a wavelength of at least 550nm and comprising at least one of NIR and IR.

10. A medical imaging method for imaging a light sensitive object (2) for imaging biological tissue (3) of an eye (4), the light sensitive object (2) having a spectral feature (36), the method comprising the steps of:

acquiring an input image (10) with an input set (15) of input spectral bands (16), the input image (10) comprising input pixels (14), the input pixels (14) comprising the input set (15),

conveying an output image (22) consisting of an output set (29) of output spectral bands (30), the output set (29) representing spectral features (36) in the output spectral bands (30) of the output set (29),

converting the input image (10) into the output image (22) by replacing at least one input-only spectral band (38) of the input set (15) with at least one output-only spectral band (39) of the output set (29) and by adapting the intensity in the at least one input-only spectral band (38) to spectral features (36) using previously stored calibration data (56).

11. The medical imaging method of claim 10, further comprising the steps of: determining a Biological Pigment (BP) in the biological tissue (3) from the input set (15) of input pixels (14) in at least one of the input images (10)₁To BP_m) And assigning the output set (29) to output pixels (26) in dependence on the local concentration.

12. Medical imaging method according to claim 10, wherein the biological tissue (3) is illuminated in a non-visible light range (18), the at least one input-only spectral band (38) being located in the non-visible light range (18), and the at least one output-only spectral band (39) being located in the visible light range (37).

13. Medical imaging method according to claim 10, wherein the illumination of the light-sensitive object (2) in the at least one output-only spectral band (39) is weaker than the illumination of the light-sensitive object (2) in the at least one input-only spectral band (38).

14. The medical imaging method according to any one of claims 10 to 13, further comprising the steps of: the output set (29) is generated from a stored representation of the model output spectral features (60).

15. A non-transitory computer storage medium storing a program that causes a computer to perform the method according to any one of claims 10 to 14.